Battery cell appearance detection method and device, storage medium and electronic equipment

Abstract

The invention discloses a battery cell appearance detection method and device, a storage medium and electronic equipment. Relates to the technical field of data processing, the field of artificial intelligence and other related technical fields. Performing feature extraction on the cylindrical surface expansion graph through a target detection neural network to obtain target feature information corresponding to the to-be-detected battery cell; and detecting the appearance of the to-be-detected battery cell based on the target feature information through the target detection neural network to obtain a target detection result corresponding to the to-be-detected battery cell. According to the method and the device, the problem that the detection efficiency of the appearance of the battery cell is relatively low due to the fact that the appearance of the coated battery cell is detected manually in related technologies is solved.

Description

Technical Field

[0001] The present application relates to the fields of data processing technology, artificial intelligence and other related technical fields, and specifically to a method and device for detecting the appearance of a battery cell, a storage medium and an electronic device. Background Art

[0002] Lithium batteries are widely used in daily life due to their high energy density, long service life and environmental protection. Finished batteries must undergo strict appearance inspection before they are put on the market. Due to the diversity, complexity and unpredictability of product appearance, lithium battery appearance defect detection has always been a difficult point in visual inspection technology. In order to prevent defective products from flowing into end customers, most of the lithium battery industry currently uses manual methods to conduct multiple inspections on their appearance. However, manual inspection is time-consuming and labor-intensive, with low efficiency and production capacity, which greatly increases the labor intensity of workers. It is also prone to false detection and missed detection, making it difficult to meet the requirements and standards of inspection and can no longer meet the needs of the market.

[0003] With regard to the problem that the appearance of the coated battery cells is manually inspected in the related art, resulting in low efficiency in the inspection of the appearance of the battery cells, no effective solution has been proposed so far. Summary of the invention

[0004] The main purpose of the present application is to provide a method and device for detecting the appearance of a battery cell, a storage medium and an electronic device, so as to solve the problem in the related art that the appearance of the battery cell after being coated is detected manually, resulting in a relatively low detection efficiency of the battery cell appearance.

[0005] In order to achieve the above-mentioned purpose, according to one aspect of the present application, a method for detecting the appearance of a battery cell is provided. The method comprises: obtaining a cylindrical expansion diagram corresponding to the battery cell to be detected; extracting features from the cylindrical expansion diagram through a target detection neural network to obtain target feature information corresponding to the battery cell to be detected; detecting the appearance of the battery cell to be detected based on the target feature information through the target detection neural network to obtain a target detection result corresponding to the battery cell to be detected, wherein the target detection result is used to characterize whether there is any abnormality in the appearance of the battery cell to be detected.

[0006] Furthermore, feature extraction is performed on the cylindrical expansion image through a target detection neural network to obtain target feature information corresponding to the battery cell to be detected, including: feature extraction is performed on the cylindrical expansion image through a feature extraction module of the target detection neural network to obtain first feature information; information aggregation is performed on the first feature information through a multi-dimensional information aggregation module of the target detection neural network to obtain second feature information; alignment conversion is performed on the second feature information through a deep feature alignment conversion module of the target detection neural network to obtain third feature information; information fusion is performed on the third feature information through a local neighborhood information fusion module of the target detection neural network to obtain the target feature information.

[0007] Further, feature extraction is performed on the cylindrical expansion image through the feature extraction module of the target detection neural network to obtain the first feature information, including: performing a convolution operation of a first scale on the cylindrical expansion image to obtain a first feature map corresponding to the cylindrical expansion image; performing a convolution operation of a second scale on the cylindrical expansion image to obtain a second feature map corresponding to the cylindrical expansion image, wherein the first scale is larger than the second scale; performing feature calculation and feature fusion on the first feature map and the second feature map to obtain a third feature map; and performing feature extraction on the third feature map to obtain the first feature information.

[0008] Furthermore, performing feature calculation and feature fusion on the first feature map and the second feature map to obtain a third feature map includes: performing an average pooling operation on the first feature map to obtain a processed first feature map; performing linear interpolation calculation on the second feature map to obtain a processed second feature map; and performing feature fusion on the processed first feature map and the processed second feature map to obtain the third feature map.

[0009] Furthermore, the first feature information is aggregated by the multi-dimensional information aggregation module of the target detection neural network to obtain the second feature information, including: performing convolution operation and linear interpolation calculation on the first feature information to obtain multiple feature components corresponding to the first feature information; aggregating the multiple feature components to obtain initial second feature information; performing convolution operation on the initial second feature information to obtain initial second feature information after convolution; and processing the initial second feature information after convolution by a target activation function to obtain the second feature information.

[0010] Furthermore, the second feature information is aligned and transformed by the deep feature alignment transformation module of the target detection neural network to obtain the third feature information, including: performing an average pooling operation on the second feature information to obtain processed second feature information; performing feature fusion on the processed second feature information to obtain fused second feature information; and performing alignment transformation on the fused second feature information to obtain the third feature information.

[0011] Furthermore, after detecting the appearance of the battery cell to be detected based on the target feature information through the target detection neural network to obtain the target detection result corresponding to the battery cell to be detected, the method also includes: if the target detection result indicates that there is an abnormality in the appearance of the battery cell to be detected, feeding back the target detection result to the target object; obtaining a target processing strategy input by the target object; and processing the battery cell to be detected according to the target processing strategy.

[0012] Furthermore, before extracting features from the cylindrical unfolded image through a target detection neural network to obtain target feature information corresponding to the battery cell to be detected, the method also includes: obtaining a training data set, wherein the training data set includes at least: multiple battery cell cylindrical unfolded images and real appearance detection results corresponding to each battery cell cylindrical unfolded image; configuring hyperparameters for the initial detection neural network to obtain a configured initial detection neural network; processing and predicting the multiple battery cell cylindrical unfolded images through the configured initial detection neural network to obtain predicted appearance detection results corresponding to the multiple battery cell cylindrical unfolded images; performing calculations based on the real appearance detection results and the predicted appearance detection results to obtain a target loss value; and updating the hyperparameters based on the target loss value to obtain the target detection neural network.

[0013] In order to achieve the above-mentioned purpose, according to another aspect of the present application, a battery cell appearance detection device is provided. The device comprises: a first acquisition unit, used to acquire a cylindrical expansion diagram corresponding to the battery cell to be detected; a feature extraction unit, used to extract features from the cylindrical expansion diagram through a target detection neural network to obtain target feature information corresponding to the battery cell to be detected; a detection unit, used to detect the appearance of the battery cell to be detected based on the target feature information through the target detection neural network to obtain a target detection result corresponding to the battery cell to be detected, wherein the target detection result is used to characterize whether there is any abnormality in the appearance of the battery cell to be detected.

[0014] Furthermore, the feature extraction unit includes: a feature extraction module, which is used to perform feature extraction on the cylindrical expansion image through the feature extraction module of the target detection neural network to obtain first feature information; an aggregation module, which is used to perform information aggregation on the first feature information through the multi-dimensional information aggregation module of the target detection neural network to obtain second feature information; a conversion module, which is used to perform alignment conversion on the second feature information through the deep feature alignment conversion module of the target detection neural network to obtain third feature information; and a fusion module, which is used to perform information fusion on the third feature information through the local neighborhood information fusion module of the target detection neural network to obtain the target feature information.

[0015] Furthermore, the feature extraction module includes: a first convolution submodule, used to perform a convolution operation of a first scale on the cylindrical expansion image to obtain a first feature map corresponding to the cylindrical expansion image; a second convolution submodule, used to perform a convolution operation of a second scale on the cylindrical expansion image to obtain a second feature map corresponding to the cylindrical expansion image, wherein the first scale is larger than the second scale; a first processing submodule, used to perform feature calculation and feature fusion on the first feature map and the second feature map to obtain a third feature map; a feature extraction submodule, used to perform feature extraction on the third feature map to obtain the first feature information.

[0016] Furthermore, the first processing sub-module includes: a pooling sub-module, used to perform an average pooling operation on the first feature map to obtain a processed first feature map; a calculation sub-module, used to perform a linear interpolation calculation on the second feature map to obtain a processed second feature map; and a fusion sub-module, used to perform feature fusion on the processed first feature map and the processed second feature map to obtain the third feature map.

[0017] Furthermore, the aggregation module includes: a second processing sub-module, used to perform convolution operation and linear interpolation calculation on the first feature information to obtain multiple feature components corresponding to the first feature information; an aggregation sub-module, used to aggregate information on the multiple feature components to obtain initial second feature information; a third convolution sub-module, used to perform convolution operation on the initial second feature information to obtain the convolved initial second feature information; and a third processing sub-module, used to process the convolved initial second feature information through a target activation function to obtain the second feature information.

[0018] Furthermore, the conversion module includes: a pooling submodule, used to perform an average pooling operation on the second feature information to obtain processed second feature information; a fusion submodule, used to perform feature fusion on the processed second feature information to obtain fused second feature information; and a conversion submodule, used to perform alignment conversion on the fused second feature information to obtain the third feature information.

[0019] Furthermore, the device also includes: a feedback unit, which is used to detect the appearance of the battery cell to be detected based on the target feature information through the target detection neural network, and after obtaining the target detection result corresponding to the battery cell to be detected, if the target detection result indicates that there is an abnormality in the appearance of the battery cell to be detected, then the target detection result is fed back to the target object; a second acquisition unit, which is used to obtain the target processing strategy input by the target object; and a first processing unit, which is used to process the battery cell to be detected according to the target processing strategy.

[0020] Furthermore, the device also includes: a second acquisition unit, which is used to acquire a training data set before extracting features of the cylindrical expansion image through a target detection neural network to obtain target feature information corresponding to the battery cell to be detected, wherein the training data set at least includes: multiple battery cell cylindrical expansion images and real appearance detection results corresponding to each battery cell cylindrical expansion image; a configuration unit, which obtains the configured initial detection neural network by configuring hyperparameters for the initial detection neural network; a second processing unit, which is used to process and predict the multiple battery cell cylindrical expansion images through the configured initial detection neural network to obtain predicted appearance detection results corresponding to the multiple battery cell cylindrical expansion images; a calculation unit, which obtains a target loss value by calculating based on the real appearance detection results and the predicted appearance detection results; and an updating unit, which obtains the target detection neural network by updating the hyperparameters based on the target loss value.

[0021] In order to achieve the above-mentioned purpose, according to one aspect of the present application, a computer-readable storage medium is provided, wherein the storage medium stores a program, wherein when the program is running, the device where the storage medium is located is controlled to execute any one of the above-mentioned battery cell appearance inspection methods.

[0022] In order to achieve the above-mentioned purpose, according to another aspect of the present application, an electronic device is also provided, the electronic device includes one or more processors and a memory, and the memory is used to store the one or more processors to implement any one of the above-mentioned battery cell appearance detection methods.

[0023] Through this application, the following steps are adopted: obtaining the cylindrical expansion diagram corresponding to the battery cell to be detected; extracting the features of the cylindrical expansion diagram through the target detection neural network to obtain the target feature information corresponding to the battery cell to be detected; detecting the appearance of the battery cell to be detected based on the target feature information through the target detection neural network to obtain the target detection result corresponding to the battery cell to be detected, wherein the target detection result is used to characterize whether there is an abnormality in the appearance of the battery cell to be detected. Through this application, the problem of low efficiency in detecting the appearance of the battery cell after coating is solved in the related art. In this scheme, the target detection neural network is first used to extract the features in the cylindrical expansion diagram corresponding to the battery cell to be detected to obtain the target feature information of the battery cell to be detected; then, based on the target feature information, the target detection neural network is used to detect the appearance of the battery cell to be detected to obtain the corresponding target detection result, and it is judged whether there is an abnormality in the appearance of the battery cell to be detected according to the target detection result. By detecting the appearance of the battery cell to be detected through the target detection neural network, the automatic detection of abnormal appearance of the battery cell is realized, and the cost of manual detection is also reduced, the detection efficiency and accuracy are improved, and it has strong scalability and versatility. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] The drawings constituting a part of the present application are used to provide a further understanding of the present application. The illustrative embodiments and descriptions of the present application are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings:

[0025] Figure 1 is a flow chart of a method for detecting the appearance of a battery cell provided in an embodiment of the present application;

[0026] Figure 2 is a schematic diagram of a feature extraction module provided according to an embodiment of the present application;

[0027] Figure 3 is a schematic diagram of a multi-dimensional information aggregation module provided according to an embodiment of the present application;

[0028] Figure 4 is a schematic diagram of a deep feature alignment conversion module provided according to an embodiment of the present application;

[0029] Figure 5 is a schematic diagram of a battery cell appearance detection method provided according to an embodiment of the present application;

[0030] Figure 6 is a schematic diagram of a battery cell appearance inspection device provided according to an embodiment of the present application;

[0031] Figure 7 is a schematic diagram of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0032] It should be noted that, in the absence of conflict, the embodiments and features in the embodiments of the present application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0033] In order to enable those skilled in the art to better understand the solution of the present application, the technical solution in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present application.

[0034] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present application described here. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0035] It should be noted that the relevant information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for display, data for analysis, etc.) involved in this disclosure are all information and data authorized by the user or fully authorized by all parties. For example, an interface is set between this system and the relevant user or organization. Before obtaining relevant information, it is necessary to send an acquisition request to the aforementioned user or organization through the interface, and obtain relevant information after receiving the consent information fed back by the aforementioned user or organization.

[0036] The present invention is described below in conjunction with preferred implementation steps. Figure 1 is a flow chart of a method for detecting the appearance of a battery cell according to an embodiment of the present application. Figure 1 As shown, the method comprises the following steps:

[0037] Step S101, obtaining a cylindrical expansion diagram corresponding to the battery cell to be tested.

[0038] Optionally, after the battery cell to be inspected is wrapped in the wrapping machine, the transport station uses appearance inspection equipment (eg, a 2.5D camera and a strip line scanning light source) to obtain a cylindrical unfolded image based on the rotation of the battery cell.

[0039] It should be noted that the 2.5D camera can combine diffuse reflection, specular reflection and low-angle reflection to extract the cylindrical features of the cylindrical battery cell (i.e. the battery cell to be tested) to obtain a cylindrical unfolded image.

[0040] Step S102, extracting features from the cylindrical expansion image through a target detection neural network to obtain target feature information corresponding to the battery cell to be detected.

[0041] Optionally, the reparameter alignment fusion module (RAF), multi-scale information aggregation module (MIA), deep feature alignment conversion module (DFAC) and local neighborhood information fusion module (LNIF) in the YoloV8 neural network model or other neural network models (i.e., target detection neural network) are used to extract features from the cylindrical expansion image to obtain target feature information corresponding to the battery cell to be detected.

[0042] Step S103, detecting the appearance of the battery cell to be detected based on the target feature information through a target detection neural network to obtain a target detection result corresponding to the battery cell to be detected, wherein the target detection result is used to characterize whether there is any abnormality in the appearance of the battery cell to be detected.

[0043] Optionally, the appearance of the battery cell to be inspected is inspected based on the target feature information of the battery cell to be inspected through a target detection neural network to obtain a target detection result (ie, whether there is any defect in the appearance of the battery cell to be inspected).

[0044] It should be noted that the appearance defects of the coated cell cylinder (i.e. the cell to be tested) include but are not limited to: pits, wrinkles, foreign matter, dirt and damage. Since the outer shell of the cylindrical cell is an aluminum shell, it is relatively light and thin, and is easily pitted by external forces. The characteristics of its defective appearance are: on the surface of the cylindrical cell, the area defects account for a large proportion and are distributed less. Since there are creases in the PET blue film of the cell, wrinkles are formed when it contacts the cylindrical cell during heat shrinkage. The characteristics of its defective appearance are: on the surface of the cylindrical cell, the wrinkles mainly occupy the entire shoulder height and present an irregular strip shape. Since the cylindrical cell is affected by the production environment during the production process, foreign matter such as blue film silk or Teflon glue will be attached during the coating; or during welding, there will be welding slag foreign matter on the surface. The characteristics of its defective appearance are: on the surface of the cylindrical cell, the area defects account for a small proportion, and the appearance is a foreign matter that is approximately circular and filamentous. Since the dirt on the PET blue film of the battery cell is mainly due to the oil stains from the oven on the surface of the blue film, the appearance of the defect is characterized by: the dirt occupies a large area on the surface of the cylindrical battery cell, the distribution is irregular, the shape is uneven, and the color difference is obvious. Since the PET blue film of the battery cell is damaged by external forces during transportation, the appearance of the defect is characterized by: the blue film is damaged to expose the aluminum shell, the color difference is obvious, it is bright white, and the shape is irregular, and there are many full welds on the cylindrical battery cell.

[0045] It should be noted that in order to improve the detection accuracy, more data sets can be used when training the YoloV8 model, and the model can be tuned and optimized. At the same time, other technologies and algorithms, including but not limited to: target tracking, image enhancement, etc., can be combined to further improve the detection effect.

[0046] In summary, the target detection neural network is first used to extract the features in the cylindrical expansion diagram corresponding to the battery cell to be detected, so as to obtain the target feature information of the battery cell to be detected; then, based on the target feature information, the target detection neural network is used to detect the appearance of the battery cell to be detected, so as to obtain the corresponding target detection result, and the target detection result is used to determine whether there is any abnormality in the appearance of the battery cell to be detected. By detecting the appearance of the battery cell to be detected through the target detection neural network, the automatic detection of abnormal appearance of the battery cell is realized, the cost of manual detection is reduced, the detection efficiency and accuracy are improved, and it has strong scalability and versatility.

[0047] Optionally, in the battery cell appearance detection method provided in the embodiment of the present application, feature extraction is performed on the cylindrical expansion image through the target detection neural network to obtain target feature information corresponding to the battery cell to be detected, including: feature extraction is performed on the cylindrical expansion image through the feature extraction module of the target detection neural network to obtain first feature information; information aggregation is performed on the first feature information through the multi-dimensional information aggregation module of the target detection neural network to obtain second feature information; alignment conversion is performed on the second feature information through the deep feature alignment conversion module of the target detection neural network to obtain third feature information; information fusion is performed on the third feature information through the local neighborhood information fusion module of the target detection neural network to obtain target feature information.

[0048] In an optional embodiment, the cylindrical expansion image is input into the YoloV8 neural network or other neural network models, and the cylindrical expansion image is subjected to feature extraction through the heavy parameter alignment fusion module (i.e., feature extraction module), and the feature information of the cylindrical expansion image is converted into a form that can be understood by the neural network to obtain the first feature information. The first feature information is aggregated by the multi-dimensional information aggregation module, and the feature information of different dimensions is combined together to extract more comprehensive and comprehensive feature information to obtain the second feature information. The second feature information is aligned and converted by the deep feature alignment conversion module, and the feature information at different levels is aligned to improve the expressiveness and discrimination of the features to obtain the third feature information. The third feature information is information-fused by the local neighborhood information fusion module, and the local information is fused with the global information to extract more discriminative features, thereby better describing the appearance characteristics of the battery cell to be tested to obtain the final target feature information.

[0049] It should be noted that the local neighborhood information fusion module deletes the process of generating feature components based on the multi-dimensional information aggregation module, and adopts a simpler local neighborhood fusion strategy. It processes large-scale features through adaptive average pooling, and fuses them with similar local neighborhood features for information fusion, so that localized feature information can be integrated with multi-layer neighborhood features.

[0050] The cylindrical expansion image is extracted through the target detection neural network, and the target feature information is obtained after a series of processing. This can improve the detection accuracy, enhance the ability to detect abnormal situations, optimize the model performance, and better describe and judge the appearance characteristics of the battery cell to be tested.

[0051] Optionally, in the battery cell appearance detection method provided in the embodiment of the present application, feature extraction is performed on the cylindrical expansion image through the feature extraction module of the target detection neural network to obtain the first feature information, including: performing a convolution operation on the cylindrical expansion image at a first scale to obtain a first feature map corresponding to the cylindrical expansion image; performing a convolution operation on the cylindrical expansion image at a second scale to obtain a second feature map corresponding to the cylindrical expansion image, wherein the first scale is greater than the second scale; performing feature calculation and feature fusion on the first feature map and the second feature map to obtain a third feature map; and performing feature extraction on the third feature map to obtain the first feature information.

[0052] In an alternative embodiment, Figure 2 is a schematic diagram of a feature extraction module provided according to an embodiment of the present application, such as Figure 2 As shown, the re-parameterized alignment and fusion module (i.e., feature extraction module) obtains defect features of different scales from the second to fourth layers and the last layer of the backbone. After receiving the cylindrical expansion image of the battery cell to be inspected, a 3×3 (i.e., first scale) convolution operation and a 1×1 (i.e., second scale) convolution operation are performed on it to obtain a large-scale feature map (i.e., first feature map) and a small-scale feature map (i.e., second feature map) of the cylindrical expansion image. The large-scale feature map and the small-scale feature map are feature-calculated and feature-fused to obtain a third feature map. The third feature map is feature-extracted through the 1×1 convolution operation, normalization operation, and SiLu activation function operation in the re-parameterized VGG structure to obtain the appearance global feature information F1 (i.e., first feature information) of the coated cylindrical battery cell (i.e., the battery cell to be inspected).

[0053] Through multi-scale feature extraction and feature fusion, we can make full use of feature information at different scales, improve the expressiveness and richness of features, and thus better describe the appearance characteristics of the battery cells to be inspected, providing a more reliable basis for subsequent anomaly detection and processing, and also improving the accuracy and robustness of target detection.

[0054] Optionally, in the battery cell appearance detection method provided in an embodiment of the present application, feature calculation and feature fusion are performed on the first feature map and the second feature map to obtain a third feature map, including: performing an average pooling operation on the first feature map to obtain a processed first feature map; performing linear interpolation calculation on the second feature map to obtain a processed second feature map; and performing feature fusion on the processed first feature map and the processed second feature map to obtain a third feature map.

[0055] In an optional embodiment, the large-scale feature map (i.e., the first feature map) is average pooled to obtain the processed first feature map; the small-scale feature map (i.e., the second feature map) is aligned to the (H / 4, W / 4) size by linear interpolation in bilinear mode to obtain the processed second feature map. The processed first feature map and the processed second feature map are concatenated and feature fused to obtain the third feature map.

[0056] By calculating, adjusting and fusing the first feature map and the second feature map to obtain the third feature map, we can make full use of multi-scale information, improve the expression ability and discrimination of the features, so as to better describe the appearance characteristics of the battery cell to be tested and provide a more reliable basis for subsequent processing and judgment.

[0057] Optionally, in the battery cell appearance detection method provided in the embodiment of the present application, the first feature information is aggregated by a multi-dimensional information aggregation module of a target detection neural network to obtain the second feature information, including: performing a convolution operation and a linear interpolation calculation on the first feature information to obtain multiple feature components corresponding to the first feature information; performing information aggregation on the multiple feature components to obtain initial second feature information; performing a convolution operation on the initial second feature information to obtain initial second feature information after convolution; and processing the initial second feature information after convolution by a target activation function to obtain second feature information.

[0058] In an alternative embodiment, Figure 3 is a schematic diagram of a multi-dimensional information aggregation module provided according to an embodiment of the present application, such as Figure 3 As shown in FIG. 1 , after receiving the first feature information F1 output by the re-parameter alignment fusion module, the multi-dimensional information aggregation module performs adaptive average pooling and 1×1 convolution operations on the first feature information to generate feature components X0 and X1, and performs bilinear interpolation calculation and 1×1 convolution operations on the first feature information to generate feature components X2. Multiple feature components X0, X1, and X2 are concatenated to unify different dimensions and fuse the local information contained in each component to obtain the initial second feature information F2. F2 is subjected to a 1×1 convolution operation to obtain the initial second feature information after convolution. The global feature component X3 (X3 is obtained by X3) corresponding to the initial second feature information after convolution is obtained through the ReLu activation function (i.e., the target activation function). l1 and X l2 The global feature component X3 and the local feature component X4 are used to calculate the second feature information F3. The calculation formula is as follows:

[0059] F3=W1X l1 +W2X4×W3X l2

[0060] Among them, F3 is the second feature information; W1 is the component X l1 The corresponding weight value; W2 is the weight value corresponding to component X4; W3 is the weight value corresponding to component X l2 The corresponding weight value.

[0061] The first feature information is aggregated through a multi-dimensional information aggregation module to obtain the second feature information, which is beneficial to improving the diversity and expression ability of the features, increasing the richness of the features, and further improving the feature description of the battery cell to be detected, thereby improving the accuracy of target detection.

[0062] Optionally, in the battery cell appearance detection method provided in an embodiment of the present application, the second feature information is aligned and converted by a deep feature alignment conversion module of the target detection neural network to obtain the third feature information, including: performing an average pooling operation on the second feature information to obtain processed second feature information; performing feature fusion on the processed second feature information to obtain fused second feature information; and performing alignment conversion on the fused second feature information to obtain the third feature information.

[0063] In an alternative embodiment, Figure 4 is a schematic diagram of a deep feature alignment conversion module provided according to an embodiment of the present application, such as Figure 4 As shown, after the deep feature alignment conversion module receives the second feature information F3 output by the multi-dimensional information aggregation module, it performs adaptive average pooling processing on it so that the feature map scale is aligned to (H / 16, W / 16) to obtain the processed second feature information F4. The deep defect features of the processed second feature information F4 are mined through the Transformer converter structure to obtain the query vector Q, key vector K and value vector V corresponding to the processed second feature information. After the query vector Q and the key vector K are fused, the Softmax smoothing result is used to fuse it with the value vector V to obtain the fused second feature information. Finally, the fused second feature information is subjected to deep feature mining and alignment conversion, and the high-level semantic information F5 (i.e., the third feature information) processed by the deep feature alignment conversion is output.

[0064] By performing average pooling, feature fusion and alignment transformation on the second feature information to obtain the third feature information, the integration and consistency of the features can be improved, the accuracy of the features can be increased, and more reliable feature information can be provided for subsequent target detection and anomaly judgment.

[0065] Optionally, in the battery cell appearance detection method provided in the embodiment of the present application, after the appearance of the battery cell to be detected is detected based on the target feature information through the target detection neural network to obtain the target detection result corresponding to the battery cell to be detected, the method also includes: if the target detection result indicates that there is an abnormality in the appearance of the battery cell to be detected, feeding back the target detection result to the target object; obtaining a target processing strategy input by the target object; and processing the battery cell to be detected according to the target processing strategy.

[0066] In an optional embodiment, based on the target detection result, it is determined whether there is any abnormality in the appearance of the battery cell to be detected. If there is an abnormality, the target detection result is fed back to the target object for subsequent processing. The target object provides a target processing strategy for the battery cell to be detected based on the received target detection result, including but not limited to: repair, replacement, scrapping and other processing methods. According to the target processing strategy provided by the target object, the battery cell to be detected is processed accordingly. For example, if the target processing strategy is repair, maintenance operations can be performed; if it is replacement, the battery cell can be replaced; if it is scrapped, the battery cell can be discarded.

[0067] Through this method, abnormal detection and subsequent processing of the battery cells to be tested can be achieved, and the abnormal battery cells can be effectively processed to ensure product quality and safety. At the same time, through the processing strategy provided by the target object, targeted processing can be carried out to improve efficiency and accuracy.

[0068] Optionally, in the battery cell appearance detection method provided in the embodiment of the present application, before performing feature extraction on the cylindrical expansion image through the target detection neural network to obtain target feature information corresponding to the battery cell to be detected, the method also includes: obtaining a training data set, wherein the training data set includes at least: multiple battery cell cylindrical expansion images and real appearance detection results corresponding to each battery cell cylindrical expansion image; configuring hyperparameters for the initial detection neural network to obtain the configured initial detection neural network; processing and predicting the multiple battery cell cylindrical expansion images through the configured initial detection neural network to obtain predicted appearance detection results corresponding to the multiple battery cell cylindrical expansion images; performing calculations based on the real appearance detection results and the predicted appearance detection results to obtain a target loss value; and updating the hyperparameters based on the target loss value to obtain the target detection neural network.

[0069] In an optional embodiment, a suitable initial detection neural network is selected, and the hyperparameters of the network are set, including but not limited to: learning rate, batch size, number of iterations, optimizer type, etc. The configured neural network is trained using a training data set. During the training process, the network will learn to extract features from the cylindrical expansion diagram of the battery cell and predict the detection results of each image. A loss function (including but not limited to: cross entropy loss function or mean square error loss function, etc.) is used to measure the difference between the predicted result (i.e., the predicted appearance detection result) and the true annotation (i.e., the true appearance detection result). The loss value of each sample is calculated, and the overall average loss value is obtained as the target loss value. According to the target loss value, an optimization algorithm (e.g., gradient descent) is used to update the hyperparameters of the network to improve the model performance. Repeat the training and verification process until a satisfactory performance indicator is achieved or a predetermined number of iterations is reached, and the current neural network is determined as the target detection neural network.

[0070] Through these steps, a target detection neural network can be constructed that can effectively extract features from the cylindrical unfolded image of the battery cell, thereby providing support for the appearance inspection of the battery cell.

[0071] It should be noted that Figure 5 Schematic diagram of a method for detecting the appearance of a battery cell according to an embodiment of the present application. Figure 5 As shown, first, the cylindrical expansion image corresponding to the rotation of the battery cell to be detected is obtained through the detection equipment. Then, the cylindrical expansion image is extracted by the feature extraction module of the target detection neural network to obtain the first feature information; the first feature information is aggregated by the multi-dimensional information aggregation module to obtain the second feature information; the second feature information is aligned and converted by the deep feature alignment conversion module to obtain the third feature information; the third feature information is fused by the local neighborhood information fusion module to obtain the target feature information. Finally, the appearance of the battery cell to be detected is detected by the target detection neural network based on the target feature information to obtain the target detection result corresponding to the battery cell to be detected.

[0072] The method for detecting the appearance of a battery cell provided in the embodiment of the present application obtains a cylindrical expansion diagram corresponding to the battery cell to be detected; extracts features from the cylindrical expansion diagram through a target detection neural network to obtain target feature information corresponding to the battery cell to be detected; detects the appearance of the battery cell to be detected based on the target feature information through the target detection neural network to obtain a target detection result corresponding to the battery cell to be detected, wherein the target detection result is used to characterize whether there is an abnormality in the appearance of the battery cell to be detected, which solves the problem that the appearance of the battery cell after encapsulation is detected manually in the related art, resulting in a relatively low detection efficiency of the appearance of the battery cell. In summary, in this scheme, firstly, the target detection neural network is used to extract features from the cylindrical expansion diagram corresponding to the battery cell to be detected to obtain the target feature information of the battery cell to be detected; then, based on the target feature information, the target detection neural network is used to detect the appearance of the battery cell to be detected to obtain the corresponding target detection result, and it is judged whether there is an abnormality in the appearance of the battery cell to be detected according to the target detection result. By detecting the appearance of the battery cell to be detected through the target detection neural network, the automatic detection of abnormal appearance of the battery cell is realized, the cost of manual detection is reduced, the detection efficiency and accuracy are improved, and it has strong scalability and versatility.

[0073] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0074] The present application also provides a battery cell appearance inspection device. It should be noted that the battery cell appearance inspection device of the present application can be used to execute the battery cell appearance inspection method provided in the present application. The battery cell appearance inspection device provided in the present application is introduced below.

[0075] Figure 6 Schematic diagram of a battery cell appearance inspection device according to an embodiment of the present application. Figure 6 As shown, the device includes: a first acquisition unit 601, a feature extraction unit 602, and a detection unit 603.

[0076] The first acquisition unit 601 is used to acquire a cylindrical expansion diagram corresponding to the battery cell to be detected;

[0077] A feature extraction unit 602 is used to extract features from the cylindrical expansion image through a target detection neural network to obtain target feature information corresponding to the battery cell to be detected;

[0078] The detection unit 603 is used to detect the appearance of the battery cell to be detected based on the target feature information through the target detection neural network, and obtain the target detection result corresponding to the battery cell to be detected, wherein the target detection result is used to characterize whether there is any abnormality in the appearance of the battery cell to be detected.

[0079] In the battery cell appearance detection device provided in the embodiment of the present application, the first acquisition unit 601 acquires the cylindrical expansion diagram corresponding to the battery cell to be detected; the feature extraction unit 602 extracts features from the cylindrical expansion diagram through the target detection neural network to obtain the target feature information corresponding to the battery cell to be detected; the detection unit 603 detects the appearance of the battery cell to be detected based on the target feature information through the target detection neural network to obtain the target detection result corresponding to the battery cell to be detected, wherein the target detection result is used to characterize whether there is an abnormality in the appearance of the battery cell to be detected, which solves the problem of low efficiency in detecting the appearance of the battery cell after coating by manual means in the related art. In this scheme, the target detection neural network is first used to extract features in the cylindrical expansion diagram corresponding to the battery cell to be detected to obtain the target feature information of the battery cell to be detected; then, based on the target feature information, the target detection neural network is used to detect the appearance of the battery cell to be detected to obtain the corresponding target detection result, and it is judged whether there is an abnormality in the appearance of the battery cell to be detected according to the target detection result. By inspecting the appearance of the battery cell to be inspected through the target detection neural network, automatic detection of abnormal battery cell appearance is achieved, which also reduces the cost of manual inspection, improves the inspection efficiency and accuracy, and has strong scalability and versatility.

[0080] Optionally, in the battery cell appearance inspection device provided in the embodiment of the present application, the feature extraction unit includes: a feature extraction module, which is used to extract features from the cylindrical expansion image through the feature extraction module of the target detection neural network to obtain first feature information; an aggregation module, which is used to aggregate the first feature information through the multi-dimensional information aggregation module of the target detection neural network to obtain second feature information; a conversion module, which is used to align and convert the second feature information through the deep feature alignment conversion module of the target detection neural network to obtain third feature information; and a fusion module, which is used to fuse the third feature information through the local neighborhood information fusion module of the target detection neural network to obtain target feature information.

[0081] Optionally, in the battery cell appearance inspection device provided in the embodiment of the present application, the feature extraction module includes: a first convolution submodule, used to perform a convolution operation on the cylindrical expansion image at a first scale to obtain a first feature map corresponding to the cylindrical expansion image; a second convolution submodule, used to perform a convolution operation on the cylindrical expansion image at a second scale to obtain a second feature map corresponding to the cylindrical expansion image, wherein the first scale is greater than the second scale; a first processing submodule, used to perform feature calculation and feature fusion on the first feature map and the second feature map to obtain a third feature map; a feature extraction submodule, used to perform feature extraction on the third feature map to obtain first feature information.

[0082] Optionally, in the battery cell appearance inspection device provided in the embodiment of the present application, the first processing sub-module includes: a pooling sub-module, used to perform an average pooling operation on the first feature map to obtain a processed first feature map; a calculation sub-module, used to perform linear interpolation calculation on the second feature map to obtain a processed second feature map; and a fusion sub-module, used to perform feature fusion on the processed first feature map and the processed second feature map to obtain a third feature map.

[0083] Optionally, in the battery cell appearance inspection device provided in the embodiment of the present application, the aggregation module includes: a second processing sub-module, used to perform convolution operation and linear interpolation calculation on the first feature information to obtain multiple feature components corresponding to the first feature information; an aggregation sub-module, used to aggregate information of the multiple feature components to obtain initial second feature information; a third convolution sub-module, used to perform convolution operation on the initial second feature information to obtain the initial second feature information after convolution; and a third processing sub-module, used to process the initial second feature information after convolution through a target activation function to obtain the second feature information.

[0084] Optionally, in the battery cell appearance inspection device provided in the embodiment of the present application, the conversion module includes: a pooling submodule, used to perform an average pooling operation on the second feature information to obtain processed second feature information; a fusion submodule, used to perform feature fusion on the processed second feature information to obtain fused second feature information; and a conversion submodule, used to perform alignment conversion on the fused second feature information to obtain third feature information.

[0085] Optionally, in the battery cell appearance detection device provided in the embodiment of the present application, the device also includes: a feedback unit, which is used to detect the appearance of the battery cell to be detected based on the target feature information through the target detection neural network, and after obtaining the target detection result corresponding to the battery cell to be detected, if the target detection result indicates that there is an abnormality in the appearance of the battery cell to be detected, then the target detection result is fed back to the target object; a second acquisition unit, which is used to obtain the target processing strategy input by the target object; and a first processing unit, which is used to process the battery cell to be detected according to the target processing strategy.

[0086] Furthermore, in the battery cell appearance detection device provided in the embodiment of the present application, the device also includes: a second acquisition unit, which is used to acquire a training data set before performing feature extraction on the cylindrical expansion image through the target detection neural network to obtain the target feature information corresponding to the battery cell to be detected, wherein the training data set includes at least: multiple battery cell cylindrical expansion images and real appearance detection results corresponding to each battery cell cylindrical expansion image; a configuration unit, which obtains the configured initial detection neural network by configuring hyperparameters for the initial detection neural network; a second processing unit, which is used to process and predict the multiple battery cell cylindrical expansion images through the configured initial detection neural network to obtain predicted appearance detection results corresponding to the multiple battery cell cylindrical expansion images; a calculation unit, which obtains a target loss value by performing calculations based on the real appearance detection results and the predicted appearance detection results; and an updating unit, which obtains the target detection neural network by updating the hyperparameters based on the target loss value.

[0087] The battery cell appearance detection device includes a processor and a memory. The above-mentioned first determination unit 601, second determination unit 602, third determination unit 603, first change unit 304, etc. are all stored in the memory as program units, and the processor executes the above-mentioned program units stored in the memory to realize accurate detection of the battery cell appearance.

[0088] The processor includes a kernel, which calls the corresponding program unit from the memory. One or more kernels can be set, and the accurate detection of the appearance of the battery cell can be achieved by adjusting the kernel parameters.

[0089] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0090] An embodiment of the present invention provides a computer-readable storage medium on which a program is stored. When the program is executed by a processor, a method for detecting the appearance of a battery cell is implemented.

[0091] An embodiment of the present invention provides a processor, which is used to run a program, wherein a battery cell appearance detection method is executed when the program is run.

[0092] like Figure 7As shown, an embodiment of the present invention provides an electronic device, the device includes a processor, a memory, and a program stored in the memory and executable on the processor, and the processor implements the following steps when executing the program: obtaining a cylindrical expansion diagram corresponding to a battery cell to be detected; performing feature extraction on the cylindrical expansion diagram through a target detection neural network to obtain target feature information corresponding to the battery cell to be detected; detecting the appearance of the battery cell to be detected based on the target feature information through a target detection neural network to obtain a target detection result corresponding to the battery cell to be detected, wherein the target detection result is used to characterize whether there is any abnormality in the appearance of the battery cell to be detected.

[0093] Optionally, performing feature extraction on the cylindrical expansion image through a target detection neural network to obtain target feature information corresponding to the battery cell to be detected includes: performing feature extraction on the cylindrical expansion image through a feature extraction module of the target detection neural network to obtain first feature information; performing information aggregation on the first feature information through a multi-dimensional information aggregation module of the target detection neural network to obtain second feature information; performing alignment conversion on the second feature information through a deep feature alignment conversion module of the target detection neural network to obtain third feature information; performing information fusion on the third feature information through a local neighborhood information fusion module of the target detection neural network to obtain target feature information.

[0094] Optionally, feature extraction is performed on the cylindrical expansion image through a feature extraction module of the target detection neural network to obtain the first feature information, including: performing a convolution operation on the cylindrical expansion image at a first scale to obtain a first feature map corresponding to the cylindrical expansion image; performing a convolution operation on the cylindrical expansion image at a second scale to obtain a second feature map corresponding to the cylindrical expansion image, wherein the first scale is greater than the second scale; performing feature calculation and feature fusion on the first feature map and the second feature map to obtain a third feature map; and performing feature extraction on the third feature map to obtain the first feature information.

[0095] Optionally, performing feature calculation and feature fusion on the first feature map and the second feature map to obtain a third feature map includes: performing an average pooling operation on the first feature map to obtain a processed first feature map; performing linear interpolation calculation on the second feature map to obtain a processed second feature map; and performing feature fusion on the processed first feature map and the processed second feature map to obtain a third feature map.

[0096] Optionally, the first feature information is aggregated by a multi-dimensional information aggregation module of the target detection neural network to obtain the second feature information, including: performing a convolution operation and a linear interpolation calculation on the first feature information to obtain multiple feature components corresponding to the first feature information; performing information aggregation on the multiple feature components to obtain initial second feature information; performing a convolution operation on the initial second feature information to obtain initial second feature information after convolution; and processing the initial second feature information after convolution by a target activation function to obtain second feature information.

[0097] Optionally, the second feature information is aligned and transformed by a deep feature alignment transformation module of the target detection neural network to obtain the third feature information, including: performing an average pooling operation on the second feature information to obtain processed second feature information; performing feature fusion on the processed second feature information to obtain fused second feature information; and performing alignment transformation on the fused second feature information to obtain the third feature information.

[0098] Optionally, after detecting the appearance of the battery cell to be detected based on the target feature information through the target detection neural network to obtain the target detection result corresponding to the battery cell to be detected, the method also includes: if the target detection result indicates that there is an abnormality in the appearance of the battery cell to be detected, feeding back the target detection result to the target object; obtaining a target processing strategy input by the target object; and processing the battery cell to be detected according to the target processing strategy.

[0099] Optionally, before extracting features from the cylindrical unfolded image through the target detection neural network to obtain target feature information corresponding to the battery cell to be detected, the method also includes: obtaining a training data set, wherein the training data set includes at least: multiple cylindrical unfolded images of battery cells and real appearance detection results corresponding to the cylindrical unfolded image of each battery cell; configuring hyperparameters for the initial detection neural network to obtain the configured initial detection neural network; processing and predicting the multiple cylindrical unfolded images of battery cells through the configured initial detection neural network to obtain predicted appearance detection results corresponding to the multiple cylindrical unfolded images of battery cells; performing calculations based on the real appearance detection results and the predicted appearance detection results to obtain a target loss value; and updating the hyperparameters based on the target loss value to obtain the target detection neural network.

[0100] The devices in this article can be servers, PCs, PADs, mobile phones, etc.

[0101] The present application also provides a computer program product, which, when executed on a data processing device, is suitable for executing a program that initializes the following method steps: obtaining a cylindrical expansion image corresponding to the battery cell to be detected; performing feature extraction on the cylindrical expansion image through a target detection neural network to obtain target feature information corresponding to the battery cell to be detected; detecting the appearance of the battery cell to be detected based on the target feature information through a target detection neural network to obtain a target detection result corresponding to the battery cell to be detected, wherein the target detection result is used to characterize whether there is any abnormality in the appearance of the battery cell to be detected.

[0102] Optionally, performing feature extraction on the cylindrical expansion image through a target detection neural network to obtain target feature information corresponding to the battery cell to be detected includes: performing feature extraction on the cylindrical expansion image through a feature extraction module of the target detection neural network to obtain first feature information; performing information aggregation on the first feature information through a multi-dimensional information aggregation module of the target detection neural network to obtain second feature information; performing alignment conversion on the second feature information through a deep feature alignment conversion module of the target detection neural network to obtain third feature information; performing information fusion on the third feature information through a local neighborhood information fusion module of the target detection neural network to obtain target feature information.

[0103] Optionally, feature extraction is performed on the cylindrical expansion image through a feature extraction module of the target detection neural network to obtain the first feature information, including: performing a convolution operation on the cylindrical expansion image at a first scale to obtain a first feature map corresponding to the cylindrical expansion image; performing a convolution operation on the cylindrical expansion image at a second scale to obtain a second feature map corresponding to the cylindrical expansion image, wherein the first scale is greater than the second scale; performing feature calculation and feature fusion on the first feature map and the second feature map to obtain a third feature map; and performing feature extraction on the third feature map to obtain the first feature information.

[0104] Optionally, performing feature calculation and feature fusion on the first feature map and the second feature map to obtain a third feature map includes: performing an average pooling operation on the first feature map to obtain a processed first feature map; performing linear interpolation calculation on the second feature map to obtain a processed second feature map; and performing feature fusion on the processed first feature map and the processed second feature map to obtain a third feature map.

[0105] Optionally, the first feature information is aggregated by a multi-dimensional information aggregation module of the target detection neural network to obtain the second feature information, including: performing a convolution operation and a linear interpolation calculation on the first feature information to obtain multiple feature components corresponding to the first feature information; performing information aggregation on the multiple feature components to obtain initial second feature information; performing a convolution operation on the initial second feature information to obtain initial second feature information after convolution; and processing the initial second feature information after convolution by a target activation function to obtain second feature information.

[0106] Optionally, the second feature information is aligned and transformed by a deep feature alignment transformation module of the target detection neural network to obtain the third feature information, including: performing an average pooling operation on the second feature information to obtain processed second feature information; performing feature fusion on the processed second feature information to obtain fused second feature information; and performing alignment transformation on the fused second feature information to obtain the third feature information.

[0107] Optionally, after detecting the appearance of the battery cell to be detected based on the target feature information through the target detection neural network to obtain the target detection result corresponding to the battery cell to be detected, the method also includes: if the target detection result indicates that there is an abnormality in the appearance of the battery cell to be detected, feeding back the target detection result to the target object; obtaining a target processing strategy input by the target object; and processing the battery cell to be detected according to the target processing strategy.

[0108] Optionally, before extracting features from the cylindrical unfolded image through the target detection neural network to obtain target feature information corresponding to the battery cell to be detected, the method also includes: obtaining a training data set, wherein the training data set includes at least: multiple cylindrical unfolded images of battery cells and real appearance detection results corresponding to the cylindrical unfolded image of each battery cell; configuring hyperparameters for the initial detection neural network to obtain the configured initial detection neural network; processing and predicting the multiple cylindrical unfolded images of battery cells through the configured initial detection neural network to obtain predicted appearance detection results corresponding to the multiple cylindrical unfolded images of battery cells; performing calculations based on the real appearance detection results and the predicted appearance detection results to obtain a target loss value; and updating the hyperparameters based on the target loss value to obtain the target detection neural network.

[0109] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.

[0110] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0111] These computer program instructions may also be stored in a computer readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture including an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0112] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process in the computer or other programmable device. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0113] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0114] The memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash RAM. The memory is an example of a computer-readable medium.

[0115] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0116] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0117] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0118] The above are only embodiments of the present application and are not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present application should be included within the scope of the claims of the present application.

Claims

1. A method for detecting the appearance of a battery cell, characterized in that: include: Obtain the cylindrical expansion diagram corresponding to the battery cell to be tested; Extracting features from the cylindrical expansion image through a target detection neural network to obtain target feature information corresponding to the battery cell to be detected; The target detection neural network detects the appearance of the battery cell to be detected based on the target feature information to obtain a target detection result corresponding to the battery cell to be detected, wherein the target detection result is used to characterize whether there is any abnormality in the appearance of the battery cell to be detected.

2. The method according to claim 1, characterized in that: The cylindrical expansion image is subjected to feature extraction by a target detection neural network, and target feature information corresponding to the battery cell to be detected is obtained, including: Performing feature extraction on the cylindrical expansion image by using a feature extraction module of the target detection neural network to obtain first feature information; Aggregating the first feature information through the multi-dimensional information aggregation module of the target detection neural network to obtain second feature information; Performing alignment conversion on the second feature information through a deep feature alignment conversion module of the target detection neural network to obtain third feature information; The third feature information is fused through the local neighborhood information fusion module of the target detection neural network to obtain the target feature information.

3. The method according to claim 2, characterized in that The feature extraction module of the target detection neural network is used to extract features from the cylindrical expansion image, and the first feature information obtained includes: Performing a convolution operation of a first scale on the cylindrical expansion image to obtain a first feature map corresponding to the cylindrical expansion image; Performing a convolution operation of a second scale on the cylinder expansion image to obtain a second feature map corresponding to the cylinder expansion image, wherein the first scale is larger than the second scale; Performing feature calculation and feature fusion on the first feature map and the second feature map to obtain a third feature map; Perform feature extraction on the third feature map to obtain the first feature information.

4. The method according to claim 3, characterized in that Performing feature calculation and feature fusion on the first feature map and the second feature map to obtain a third feature map includes: Performing an average pooling operation on the first feature map to obtain a processed first feature map; Performing linear interpolation calculation on the second feature map to obtain a processed second feature map; Feature fusion is performed on the processed first feature map and the processed second feature map to obtain the third feature map.

5. The method according to claim 2, characterized in that: The first feature information is aggregated by the multi-dimensional information aggregation module of the target detection neural network to obtain the second feature information, including: Performing a convolution operation and a linear interpolation calculation on the first feature information to obtain a plurality of feature components corresponding to the first feature information; Aggregating the multiple feature components to obtain initial second feature information; Performing a convolution operation on the initial second feature information to obtain the convolved initial second feature information; The initial second feature information after the convolution is processed by a target activation function to obtain the second feature information.

6. The method according to claim 2, characterized in that The second feature information is aligned and converted by the deep feature alignment conversion module of the target detection neural network to obtain the third feature information including: Performing an average pooling operation on the second feature information to obtain processed second feature information; Performing feature fusion on the processed second feature information to obtain fused second feature information; Performing alignment transformation on the fused second feature information to obtain the third feature information.

7. The method according to claim 1, characterized in that After detecting the appearance of the battery cell to be detected based on the target feature information by the target detection neural network to obtain the target detection result corresponding to the battery cell to be detected, the method further includes: If the target detection result indicates that the appearance of the battery cell to be detected is abnormal, the target detection result is fed back to the target object; Obtaining a target processing strategy for the target object input; The battery cell to be detected is processed according to the target processing strategy.

8. The method according to claim 1, characterized in that Before extracting features from the cylindrical expansion image through a target detection neural network to obtain target feature information corresponding to the battery cell to be detected, the method further includes: Acquire a training data set, wherein the training data set at least includes: a plurality of battery cell cylindrical expansion images and a real appearance detection result corresponding to each battery cell cylindrical expansion image; configuring hyperparameters for the initial detection neural network to obtain the configured initial detection neural network; Processing and predicting the plurality of battery cell cylindrical surface development images through the configured initial detection neural network to obtain predicted appearance detection results corresponding to the plurality of battery cell cylindrical surface development images; Calculating according to the actual appearance detection result and the predicted appearance detection result to obtain a target loss value; The hyperparameters are updated according to the target loss value to obtain the target detection neural network.

9. A computer-readable storage medium, characterized in that: The computer-readable storage medium includes a stored program, wherein when the program is executed, the storage medium is controlled to execute the battery cell appearance detection method according to any one of claims 1 to 8 on a device.

10. An electronic device, characterized in that: It includes one or more processors and a memory, wherein the memory is used to store one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the battery cell appearance detection method according to any one of claims 1 to 8.

Images